Using unmanned aerial vehicles (UAVs) to gather wide-area environmental information and extract valuable road data, we can provide reliable navigational data for unmanned ground vehicles (UGVs), which will help them drive autonomously. This collaboration between UAVs and UGVs complements each other, creating an efficient and multifunctional multi-unmanned system. The task of road extraction can be broken down into three subtasks: extracting the binary image of the road, extracting the road network graphs, and extracting key waypoints. These tasks require three types of operations: image segmentation, image thinning, and corner detection. Many innovative works have been carried out in these areas, which have been analyzed and summarized in this article. A three-stage process (BNW) has been proposed, considering factors such as the level of abstraction of the road extraction results and memory usage. The paper also discusses some issues and future trends related to road extraction tasks.